Add regularization terms in klqp.py

[siddharth-agrawal]

Added regularization terms in the loss functions and gradient calculations in klqp.py. This partially fixes #529. If this change is on the right track, will update the remaining files in the next few commits.

[dustinvtran]

Thanks!

I'm okay with not having this include a unit test. Perhaps if or when you implement it for ed.MAP, it would be nice to show they return the same answer.

Can you add a sentence or two in the Notes section of each KLqp class? E.g.,

The objective function also adds to itself a summation over all tensors 
in the `REGULARIZATION_LOSSES` collection.

[siddharth-agrawal]

@dustinvtran Updated with the relevant changes.

[dustinvtran]

Thanks!